{"schema_version":"0.10.0","id":"LR-2026-0735","record_version":1,"title":"Hacktron researchers use Claude to access OpenAI employee ChatGPT accounts","verification":{"human_verified":false,"method":"automated_gate","auto_verified":true,"score":49.8,"score_breakdown":{"source_authority":12.1,"independent_corroboration":7.0,"implicated_party_confirmed":0.0,"citation_integrity":14.0,"specificity":6.0,"completeness":3.2,"unhedged_language":5.0},"checks":{"schema_valid":true,"every_field_cited":true,"quotes_verbatim":true,"source_fetched":true,"distinct_incident":true,"dates_plausible":true,"sources_agree":true,"source_describes_incident":true,"severe_backed_by_source":true,"record_integrity":true,"eligible":true,"about_this_case":true},"gate_version":"gate-2026-10-18.7","cross_check_model":null,"cross_check_agreement":null,"duplicate_of":null,"verified_by":null,"verified_at":"2026-11-01","extraction_run_id":null,"evidence_level":{"level":"preliminary","reason":"1 report: Newscord · subject to change","original_document":null,"reports_total":1,"reports_independent":1,"reports_derivative":0,"developing":false,"disputed":false,"rules_version":"evidence-2026-10-29.1","computed_at":"2026-11-01","history":[{"date":"2026-11-01","from_level":null,"to_level":"preliminary","reason":"1 report: Newscord · subject to change"}]},"sources_hash":"0aa8a905ccc391ac23d372d32211872dc0fef339cc687a80bf42e90129a62948"},"system":{"model_name":"Claude","model_version":null,"developer":"Anthropic","harness":null,"tools_granted":[],"permissions_scope":null,"autonomy_level":"unknown","guardrails_present":[],"deployment_context":"unknown","tier":null,"lifecycle_stage":null,"integration_type":null,"intended_use":null,"safeguard_state":null,"production_intended":null},"event":{"date_occurred":null,"date_disclosed":"2026-09-18","incident_class":"security_compromise","failure_class":null,"agent_involved":null,"secondary_failure_classes":[],"trigger":"unknown","duration":null,"detection_method":"unknown","summary":"Hacktron AI used Anthropic's Claude to access OpenAI employee ChatGPT accounts and internal code, reached via an OpenAI staff discussion forum hosted on the Discourse platform. The access allowed reading private software and proposing changes to OpenAI's code repositories. The researchers said they showed permission levels by making a harmless pull request rather than downloading code. The intrusion reportedly began on July 23 with a flaw in how Discourse processed certain image files, and OpenAI said it had found two problems, one in Discourse and one in OpenAI itself. OpenAI said it had resolved both, constrained login token permissions for Community, and revoked affected tokens and sessions. The researchers were paid $6,500 under OpenAI's bug bounty program.","tags":[],"date_detected":null,"near_miss":null,"root_cause":null,"root_cause_detail":"a flaw in how Discourse processed certain image files","failure_class_detail":null,"responsible_party":null,"adversarial_actor":null,"input_provenance":null,"time_to_contain":null,"operations_paused":null,"remediation":null,"precursor_events":null},"impact":{"direct_loss_usd":null,"remediation_hours":null,"records_exposed":null,"third_parties_affected_count":null,"severity":"moderate","severity_because":"Test: data and systems of others touched (OpenAI employee accounts and code). Scope is not stated as bounded or unbounded; the only scope sentence is a researcher's theoretical claim, so the moderate tier is a judgement.","harm_class":[],"affected_description":"OpenAI employee ChatGPT accounts and internal code","protected_class_involved":null,"data_categories_exposed":[],"loss_range_usd":null,"reversibility":null,"downstream_exposure_count":null},"evidence":{"sources":[{"url":"https://newscord.org/article/hacktron-ai-researchers-use-anthropics-claude-to-hack-openai-access-chatgpt-acco--Story_20260918_ResearchersusedClaud2b04bbd6","publisher":"NewsCord","date":"2026-09-18","type":"press","title":"Hacktron AI Researchers Use Anthropic’s Claude To Hack OpenAI, Access ChatGPT Account: how 19 outlets framed it - NewsCord","archived_url":null,"attachment":"first_report","independence":"independent"}],"per_field_confidence":{"system.model_name":"C","system.developer":"C","event.incident_class":"C","event.root_cause_detail":"C","impact.affected_description":"C","impact.severity":"D","impact.severity_because":"D","event.summary":"C","system.autonomy_level":"D","system.deployment_context":"D","event.trigger":"D","event.detection_method":"D","context.country":"D","event.date_disclosed":"D","context.denominator_available":"D"},"field_citations":{"system.model_name":[{"source_index":0,"quote":"Hacktron AI used Anthropic's Claude to access OpenAI employee ChatGPT accounts and internal code."}],"system.developer":[{"source_index":0,"quote":"Hacktron AI used Anthropic's Claude to access OpenAI employee ChatGPT accounts and internal code."}],"event.incident_class":[{"source_index":0,"quote":"Hacktron AI used Anthropic's Claude to access OpenAI employee ChatGPT accounts and internal code."}],"event.root_cause_detail":[{"source_index":0,"quote":"El Mundo said the intrusion began on July 23, when Hacktron researchers found a flaw in how Discourse processed certain image files, and that OpenAI stated it had discovered two problems—one in Discourse and another in OpenAI itself."}],"impact.affected_description":[{"source_index":0,"quote":"Hacktron AI used Anthropic's Claude to access OpenAI employee ChatGPT accounts and internal code."}],"impact.severity":[{"source_index":0,"quote":"Hacktron AI used Anthropic's Claude to access OpenAI employee ChatGPT accounts and internal code."}],"impact.severity_because":[{"source_index":0,"quote":"Hacktron AI used Anthropic's Claude to access OpenAI employee ChatGPT accounts and internal code."}],"event.summary":[{"source_index":0,"quote":"Hacktron AI used Anthropic's Claude to access OpenAI employee ChatGPT accounts and internal code."},{"source_index":0,"quote":"Access enabled reading private software and proposing changes to OpenAI's code repositories."},{"source_index":0,"quote":"The Guardian described the operation as an “ethically hacked” breach in which Hacktron used Claude to access ChatGPT accounts via an OpenAI staff discussion forum hosted by the Discourse platform, then made a harmless pull request to OpenAI’s GitHub software repository."},{"source_index":0,"quote":"El Mundo said the intrusion began on July 23, when Hacktron researchers found a flaw in how Discourse processed certain image files, and that OpenAI stated it had discovered two problems—one in Discourse and another in OpenAI itself."},{"source_index":0,"quote":"El Mundo added that OpenAI said it had resolved both issues and that it constrained the login token permissions for Community and revoked the affected tokens and sessions."},{"source_index":0,"quote":"Ars Technica reported that the researchers were paid $6,500 by OpenAI as part of a bug bounty program, and that they had been given access to an Anthropic tool specifically designed for security professionals."}]},"notes_on_gaps":"Only one source was given: an automated press aggregator that cites other outlets. It is not a primary report from OpenAI, Hacktron or Anthropic, and the claim that 19 outlets reported it the same way is unverified. The year of the July 23 and July 28 dates is not stated, so date_occurred is omitted. No date of disclosure or detection is stated, so those are omitted. The sources do not say whether Claude acted autonomously; it is described as a tool used by people, so agent_involved, failure_class and the agent-related fields are omitted. No trigger or detection method is stated, so those are omitted. Root cause 'infrastructure' is my judgement that the Discourse image-processing flaw is the failed component; OpenAI's own flaw is not described. Control A3 'failed' is my inference from OpenAI constraining token permissions and revoking tokens. Discourse's 22:49:45 UTC fix time is relative to an unnamed earlier point, so no full date is given. The 'more than 1,000 OpenAI agents' Hugging Face incident is a separate event and is not reported. Financial loss is not stated (the $6,500 is a bounty paid to researchers), so direct_loss is omitted. Severity 'moderate' is my judgement: others' systems were touched, but the bounded scope is only implied by the harmless pull request. Reversibility, records exposed and third-party counts are not stated and are omitted. The OpenAI security audit and staff reassignment reported by Babnet are uncorroborated and not recorded as remediations. [case created by the backfill: read by claude-haiku-5-5, backfill-2026-11-01.2, 2026-11-01; not yet reviewed] [2026-11-01: evidence level set: Preliminary (1 report: Newscord · subject to change)]"},"context":{"industry":null,"org_size_band":null,"country":"unknown","denominator_available":false,"denominator":null,"related_incident_ids":[],"systemic_exposure":null,"shared_model_id":null,"upstream_incident_id":null,"incident_evidence":null,"regulatory":null,"coverage":null},"narrative":{"steps":[{"step":"doing","text":"Hacktron researchers used Anthropic's Claude to access OpenAI employee ChatGPT accounts and internal code.","quote":"Hacktron AI used Anthropic's Claude to access OpenAI employee ChatGPT accounts and internal code.","source_index":0,"grade":"C"},{"step":"went_wrong","text":"The intrusion reportedly began with a flaw in how Discourse processed certain image files, and OpenAI said it had found two problems.","quote":"El Mundo said the intrusion began on July 23, when Hacktron researchers found a flaw in how Discourse processed certain image files, and that OpenAI stated it had discovered two problems—one in Discourse and another in OpenAI itself.","source_index":0,"grade":"C"},{"step":"reached","text":"The researchers reached ChatGPT accounts through an OpenAI staff forum and made a harmless pull request to OpenAI's GitHub repository.","quote":"The Guardian described the operation as an “ethically hacked” breach in which Hacktron used Claude to access ChatGPT accounts via an OpenAI staff discussion forum hosted by the Discourse platform, then made a harmless pull request to OpenAI’s GitHub software repository.","source_index":0,"grade":"C"},{"step":"done","text":"OpenAI said it resolved both issues, constrained login token permissions for Community, and revoked affected tokens and sessions.","quote":"El Mundo added that OpenAI said it had resolved both issues and that it constrained the login token permissions for Community and revoked the affected tokens and sessions.","source_index":0,"grade":"C"}],"model":"claude-haiku-5-5","prompt_version":"backfill-2026-11-01.2","extracted_at":"2026-11-01","source_urls":["https://newscord.org/article/hacktron-ai-researchers-use-anthropics-claude-to-hack-openai-access-chatgpt-acco--Story_20260918_ResearchersusedClaud2b04bbd6"]},"analysis":{"sections":{"timeline":[{"text":"The intrusion began when Hacktron researchers found a flaw in how Discourse processed certain image files.","quote":"El Mundo said the intrusion began on July 23, when Hacktron researchers found a flaw in how Discourse processed certain image files, and that OpenAI stated it had discovered two problems—one in Discourse and another in OpenAI itself.","source_index":0,"when":"July 23"},{"text":"Discourse published advisory GHSA-vhm9-85gw-x335.","quote":"CyberSecurityNews said OpenAI confirmed a fix at 22:49:45 UTC, roughly 14 hours later, while also describing that Discourse published advisory GHSA-vhm9-85gw-x335 on July 28 and that OpenAI later awarded $6,500 for the OpenAI-side finding.","source_index":0,"when":"July 28"},{"text":"OpenAI confirmed a fix, roughly 14 hours later than an earlier point the source does not name.","quote":"CyberSecurityNews said OpenAI confirmed a fix at 22:49:45 UTC, roughly 14 hours later, while also describing that Discourse published advisory GHSA-vhm9-85gw-x335 on July 28 and that OpenAI later awarded $6,500 for the OpenAI-side finding.","source_index":0,"when":"22:49:45 UTC"},{"text":"OpenAI later awarded $6,500 for the OpenAI-side finding.","quote":"CyberSecurityNews said OpenAI confirmed a fix at 22:49:45 UTC, roughly 14 hours later, while also describing that Discourse published advisory GHSA-vhm9-85gw-x335 on July 28 and that OpenAI later awarded $6,500 for the OpenAI-side finding.","source_index":0,"when":null}]},"model":"claude-haiku-5-5","prompt_version":"backfill-2026-11-01.2","extracted_at":"2026-11-01","source_urls":["https://newscord.org/article/hacktron-ai-researchers-use-anthropics-claude-to-hack-openai-access-chatgpt-acco--Story_20260918_ResearchersusedClaud2b04bbd6"]}}